5th New England Go Open - Rating Changes

The AGA rated the tournament on September 7. Comparing the TDList either side of that update: 78 players with games, 189 games, and 19 players at a different rank than they started.

Tournament report · Round 5 cross-tab · Tournament page · Every player's numbers (Google Sheet) · Method & code


Eleven promotions

PlayerW-LRankChange
Aris Tian4-17k → 5k+2.40936
Jason Qi4-13k → 2k+1.34696
Luca Bove1-25k → 4k+1.20543
Binyi Lu4-114k → 13k+1.06417
Harvey Yu4-13k → 2k+0.84780
Andrew Perry4-117k → 16k+0.52846
Mateo Velikonja4-11k → 1d+0.36977
Andrew Huang4-15d → 6d+0.33989
Ryan Horner4-15k → 4k+0.16871
John Kerpan3-21d → 2d+0.15085
Micah Steinberg4-14k → 3k+0.06136

Micah Steinberg got the most out of the least: +0.06136 was the smallest gain of any four-win player here, and it was enough because he started five hundredths of a point from the line. He is now a 3k by eight thousandths, the second-weakest of the country's 149 active* 3k players.

Forest Song moved least of anyone who played - five games, 3-2, and a change of +0.00201. That is what a correctly rated player looks like.

Eight demotions

Eight players lost a rank. Five of them went 1-4 and one went 0-5; two dropped two ranks at once. Across everyone who lost points, the average was -0.27.

One demotion is worth naming, because it is the strangest line in the file: Jeremy Chiu went 4-1, won the tournament, and lost a stone. 8.03208 → 7.98574, a change of -0.04634 - and 0.032 was all the 8d interval he had left. He went in the third-weakest of the country's 26 active* 8d players; he comes out the strongest of the 52 active 7d.

Did the weekend's results predict the rating change?

Monotonically, yes: each additional win moves the column average up, from -1.84 at no wins to +0.65 at four. But the spread inside a column is wider than the gap between columns. Who you played matters as much as how often you won.

gained rating    lost rating    column mean

The 62 players who played all five rounds, one dot each. Vertical position is the rating change with the kyu/dan gap closed, so one unit is one rank. Nobody went 5-0. Pearson r = 0.71.

First-time players

Nine players arrived with no AGA rating, so their entry rank was an estimate. The rating that came out is the verdict on it - five of the nine landed within one rank of the guess.

Entered asW-L Rating assignedRankvs. guess
5D2-35.197585dspot on
3K1-4-4.311684k-1
9K4-1-8.429008k+1
8K1-4-10.0411310k-2
18K4-1-16.4625616k+2
15K1-4-17.3533017k-2
19K3-2-18.8148818k+1
19K2-3-20.9842720k-1
25K0-5-27.9498027k-2